Anal. Chem. 1990, 62, 349-353
should behave in a similar fashion (19).One way to compare the uniformity of liquid films would be through the measured plate numbers. Uniform films should produce more efficient chromatographic behavior. However, although more plates were generated in the hydrocarbon solvent systems, since k’ values and flow rates were different, a meaningful comparison cannot be made. It is not obvious either what the effect of nonwetting would be on the derived activity coefficients, or how to circumvent the problem if the effect is detrimental to their accuracy. One could start with a fused silica column containing a cross-linked polar phase such as a polyethylene glycol or a propylcyanosiloxane, which should be wet by polar solvents such as water, but these phases would of course change solute retention behavior and obfuscate interpretation of the data. This problem is currently under investigation by us. ACKNOWLEDGMENT We thank C. Wysocki of B P Research for technical assistance.
349
(4) Locke, D. C. Adv. Chromatogr. 1976, 74, 87. (5) Conder, J. R.; Young, C. L. Physicochemical Measurements by Gas Chromatography; Wiley: New York, 1979. (6) Laub, R. J.; Pecsok, R. L. Physicochemical Applications of Gas Chromatography; Wiley: New York, 1978. (7) Thomas, E. R.; Newman, B. A.; Long, T. C.; Wood, D. A.; Eckert, C. A. J. Chem. Eng. Data 1982, 27, 399. (8) Eckert, C. A.; Newman, B. A.; Nicolaides, G. L.; Long, T. C. AIChE J. 1981, 27, 33. (9) Terasawa, S.; Itsuki, H.; Yamaki, H. Anal. Chem. 1988, 58, 3021. (10) Itsuki, H.; Terasawa, S.; Yamana, N.; Ohotaka, S. Anal. Chem. 1987, 5 9 , 2916. (11) Craven, J. S.; Clauser. D. E. Analusis 1980, 8(1), 1. (12) American Instiiute of Chemical Engineers, Design Instiiute for Physical Property Data. Project 807, DCAP I I Users Guide; The Pennsylvania State University: University Park, PA, 1983. (13) Ambrose, D.; Lawrenson, I.J. J. Chem. Thermodyn. 1972, 4 , 755. (14) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Prope&s of Gases and Liquids, 4th ed.; McGraw-Hill: New York, 1987. (15) Banerjee, S. Environ. Sci. Technoi. 1985, 79, 369. (16) Littlewood, A. B. Gas Chromatography; Academic Press: New York, 1967; p 127. (17) Jennings, W. Gas Chromatography with Glass Capillary Columns, 2nd ed.; Academic Press: New York. 1980; p 21. (18) Grob, K. Helv. Chim. Acta 1965. 48, 1362. (19) Jennings, W. Gas Chromatography with Glass Capillary Columns, 2nd ed.; Academic Press: New York, 1980; p 33.
LITERATURE CITED (1) Belfer, A. I.Neftekhimiya 1972, 72, 435; Chem. Abstr. 1973, 78, 20591. (2) Belfer, A. J.; Locke, D. C. Anal. Chem. 1984, 5 6 , 2465. (3) Yang, Y.; Xiao, S.; Li, H.; Fu, Y. J. Chengdu Univ. Sci. Technol. 1988, No. 1, 35; Chem. Abstr. 1988, 709, 157451.
RECEIVED for review September 1,1989. Accepted November 22, 1989. This work was supported in part a t QC by grants from the National Science Foundation (CHE-8420326) and the PSC-CUNY FRAP Program.
Pulse Voltammetric Techniques at Microelectrodes in Pure Solvents Malgorzata Ciszkowska and Zbigniew Stojek
Department of Chemistry, Warsaw University, ul. Pasteura 1, 02-093 Warsaw, Poland Janet Osteryoung*
Department of Chemistry, State University of New York, University at Buffalo, Buffalo, New York 14214
Experlmentai conditions are described for application of pulse voltammetric techniques with microelectrodes in solvents Containing no deliberately added supporting electrolyte. Elimination of Supporting electrolyte was found to be advantageous In the case of alkyl Iodides, for which supportlng electrolyte strongly Influences the voltammetric curves. The primary reductbn waves obtained at mercury mkroelectrodes were free from phenomena due to adsorption and following chemical reaction. The heights of both linear scan reduction and reverse pulse oxidation curves were linearly dependent on ethyl, butyl, and decyl iodide concentration.
Research on microelectrodes is focused on the ways of preparation, properties, theory, and application of these electrodes (I, 2). One of the important properties of microelectrodes is a very low level of current flowing through the electrochemical cell. Consequently, the ohmic dorp (iR) may be very small despite large resistance of the solution. Therefore it is possible, with the proper instrumentation, to monitor electroactive species in solvents containing virtually no added supporting electrolyte (3-8). These conditions can be interesting, since any salt used as the supporting electrolyte can be a source of unwanted impurities in trace analysis and
a source of unwanted water in nonaqueous solvents. The first published papers on this subject proved that obtaining a curve in a pure solvent and measuring its height are possible. However, many bothersome questions remain. First of all, how does ion migration current contribute to the total current? For large-area electrodes it is well-known that at low supporting electrolyte concentrations the reduction signal of positively charged species can be appreciably higher than that at high concentrations. An experimental study done with microelectrodes showed minor changes of the limiting reduction current for Fe(CN)6-3with a change in supporting electrolyte concentration in water (6). This result is linked to substantially enhanced diffusional transport to electrodes of very small area. On the other hand, it is not surprising that uncharged molecules give wave heights virtually independent of the electrolyte level ( 3 , 4 ,6). A recent paper by Oldham deals with theory of microelectrode steady-state voltammetry with various ratios of reactant to supporting electrolyte (9). At a hemispherical electrode, for reactions that engender an increase in ionic strength a t the microelectrode interface, the faradaic redistribution of ions is predicted to diminish the ohmic overvoltage. Simultaneously the counterions will be brought into the neighborhood of the microelectrode so effectively that a very low electrolyte concentration can behave as “excess” supporting electrolyte. In the present case of no
0003-2700/90/0362-0349$02.50/00 1990 American Chemical Society
350
ANALYTICAL CHEMISTRY, VOL. 62, NO. 4, FEBRUARY 15, 1990
deliberately added electrolyte, local departure from electroneutrality may shift the autodissociation reaction of the solvent to the right (10). Another problem is choosing the optimal scheme for potential control and current measurement. Since the currents are very small, a two-electrode system can be used (4,8). In such a system all electrodes exhibit the same ohmic potential drop at steady state (11). Such a simple electrical circuit can be employed for amperometry (12,13) or other simple electroanalytical techniques. A pulse voltammetric technique in a solvent containing no supporting electrolyte would seem to be much more demanding. First, it is convenient to use a commercial instrument based on a high input impedance potentiostat; second, one has to take into account the cell time constant, to make sure the working electrode has been polarized to the desired potential during the (usually short) pulse time. There are apparently only two reports of pulse techniques applied in a pure solvent, square wave voltammetry of ethylenediaminetetraacetic acid in pure water (7) and differential pulse voltammetry of ferrocene in acetonitrile (14). A special reason for abandoning addition of a supporting electrolyte is strong interaction of the electrolyte with the substrates or products of the electrode reaction, which complicates the interpretation of the results. This is a common circumstance, particularly in the study of organic molecules. We have previously examined the reduction of several alkyl iodides at mercury electrodes under conventional voltammetric conditions ( 1 5 ) . The mechanism of the reduction was investigated by employing reverse pulse voltammetry. This provides a particularly challenging case, for the potential range of interest requires pulse amplitudes of up to 3 V. Thus the test compounds for these investigations were chosen to be decyl, butyl, and ethyl iodide. Numerous papers have been devoted to electrochemistry of alkyl halides. Those dealing with alkyl iodides at mercury electrodes have been reviewed critically in ref 15 and the references therein. Alkyl iodides are known to produce poorly defined staircase and normal pulse polarographic waves. A large maximum generally appears on the plateau of the waves, and the half-wave potentials shift toward positive potentials with a decrease in the size of the cation in tetraalkylammonium supporting electrolytes. The two-electron process consists of one or two steps, depending on the supporting electrolyte used. Although the electrochemistry of this class of compounds has been studied extensively, some aspects of the electrode process still are not clear. Mercury electrodeposited on a platinum disk was chosen as the working electrode in the present work, as we had used this electrode successfully previously in aqueous solution without supporting electrolyte (7). Our first specific objective was to identify solvents that would yield reproducible results. Second, we wished to see if the mercury microelectrode would behave reasonably with large pulse amplitudes in nonaqueous solvent. Third, we wished to test the practical range of pulse amplitude, pulse duration, and analyte concentration over which quantitative results without obvious distortions could be obtained. Finally, recognizing deficits in basic physical understanding of such systems, we wished to see what quantitative results could be obtained from these preliminary experiments. EXPERIMENTAL SECTION Normal pulse, reverse pulse, and linear scan voltammetric measurements (NPV, RPV, LSV) were carried out with a three-electrode system. Mercury deposited on a platinum microdisk, a saturated calomel electrode (SCE),and a platinum wire served as the working, reference, and auxiliary electrodes, respectively. Bare platinum and gold microdisk electrodes were also used in preliminary experiments. The SCE was separated from the cell by a bridge containing pure solvent. A two-electrode
system was also used in some experiments. A number of instruments were used,but mostly a Laboratorni Pristroje P-03 pulse polarograph, a PARC 173 potentiostat, and a BAS-100 electrochemical analyzer. The measurements of very low currents were done by connecting a Keithley 427 current amplifier to the electrochemical instrumentation by the way described in either ref 7 or ref 16. An IBM/XT Turbo computer (10 MHz) served as a storage device for the purpose of background subtraction. Background curves were obtained in the absence of analyte and subtracted point-by-point. The cathodic waves for the alkyl iodides are not well-separated from the solvent reduction. In the worst case the background current was as large as 50% of the current for reduction of the alkyl iodide. Two platinum microdisk electrodes of radius r = 12.5 and 1 pm were employed. The platinum wires were sealed into glass capillaries according to published procedures (2). Mercury was plated from 0.01 M Hg(1) solution at -0.5 V vs SCE. To control the amount of mercury on the platinum surface, the charge required to oxidize the mercury was determined after each series of experiments. Usually, the amount of mercury plated on the platinum microdisk (r = 12.5 pm) was 48 pC, which corresponds to 0.49 nmol. (Note that the amount of mercury must be large enough so that the amount removed by anodization is negligible.) We assume this electrode is a segment of a sphere, the radius of the planar intersection being the radius, r, of the platinum substrate. The volume of mercury corresponding to 48 pC is 7.34 X cm3. The volume of the segment of a sphere of radius R = (h2+ r 2 ) / 2 his V = nh(3r2+ h 2 ) / 6 ,where h is the height of the segment. Comparison with the experimental volume yields R = 13.3 X cm and h = 17.84 X cm. The area of this spherical segment is A = r ( h 2 + r2) = 1.49 X low5cm2. It is necessary to deal with this geometry rather than the simple hemisphere because the spherical segment with h > r is more stable than that with h = r and because it is experimentally impractical to deposit quantitatively each time the amount of mercury equivalent to a hemisphere. It has been shown empirically that the diffusion-limited steady-state current at such an electrode is described by ,i = knFDCr where k is an empirical constant, a function of h / r that has the value 4 when h = 0 and 27r when h = r (hemisphere) (17). The time dependence of the current is given approximately by
i = ( n F A D C / r ) ( 2 / ( 7 r ~ )+' / ~l),
T
I 1
(1)
i = knFDCr(1 + k / 7 r 3 / 2 ~ 1 / 2 ) , T >- 1 (2) where T = at/?. These equations are correct for the hemisphere and correct in first order for a disk. Thus they should be a reasonable approximation for the spherical segment. For r = 12.5 pm and h = 17.84 pm, k = 8.245 (17). Before the Hg/Pt electrode was inserted into the cell, it was dried by washing with ethyl alcohol and acetone. Mercury electrodes based on a platinum substrate can cause problems in cases where certain metal amalgams are formed, due to formation of intermetallic compounds. However, for the purposes of this paper, which does not involve any metal deposition, such electrodes appear to function reliably (7). Propylene carbonate (Fluka AG) was distilled under vacuum before use. Alkyl iodides were stabilized with metallic copper. RESULTS AND DISCUSSION Preliminary Experiments. Certainly, not all solvents can be used in electrochemical experiments without supporting electrolyte. There is a limit, which can be expressed in terms of dielectric constant, e, below which a certain amount of supporting electrolyte is necessary to measure a current response. This limit depends on the electrode area (4, 6, 18) and should also depend on the type of the instrument used. In these terms electrochemisty without supporting electrolyte has its roots in the measurements in highly resistive media of Lines and Parker, who examined some aromatic compounds in benzene and chlorobenzene solutions also containing tetrahexylammonium salts (19). Water-dioxane mixtures can cover a wide range of e, so this medium was chosen in this work to determine the experimental limits in pure solvents for the particular electro-
ANALYTICAL CHEMISTRY, VOL. 62, NO. 4, FEBRUARY 15, 1990 I
351
I
3-
t0.3
2-
t0.4 70
1-
t0.5
60
0-
+0.6 1
I
I
I
10
20
30
40
E
Figure 1. E,,, (triangles),,i (squares),and slope (circles)of LSV curves plotted vs dielectric constant of water-dioxane mixtures: 0.5 mM ferrocene; Pt microdisk radius, 12.5 pm (empty marks) and 1 pm (filled marks); T = 25 'C.
chemical instrumentation used. Ferrocene served as the depolarizer. Examining , , i Ellz,and the slope, log ((id - i ) / i d ) , of the ferrocene wave, we have found that a 12.5-pm electrode can be used in solvents o f t not lower than 10. For t greater than 30 all three measured quantitites changed negligibly. In the range 10 < e < 30, the limiting current was still constant, but ElI2shifted toward positive potentials and the slope increased with decreasing e, as illustrated in Figure 1. Substituting a 1-pm Pt disk for a 12.5-pm electrode widened the useful range of t. These results agree with the earlier results of Bond, Fleischmann, and co-workers ( 4 , 18), who applied electrodes of radius even smaller than 1 pm. Based on these experiments, propylene carbonate (PC), a nontoxic liquid with e = 64.9, appears to be an excellent solvent, and it was used in the following investigations. It should be emphasized that these comments apply only for a potentiostat with a suitably high input impedance electrometer (Z > 10" 8, e.g., PA 03, PARC l73,174A, and 273). The measurements are impossible otherwise. Linear Scan Voltammetry. Linear scan voltammetric curves obtained in pure propylene carbonate containing decyl iodide are presented in Figure 2. Since the curves are close to the cathodic limit of the potential window, the background had to be subtracted. We assume the dimensionless time for a linear scan experiment is T = D/nfv?, where u is scan rate and f = F/RT = 38.9 V-l. For the conditions of Figure 2, the non-steady-state contribution to the maximum current is approximately 19%, based on theory for a disk (20). The reduction wave is split, but it exhibits no maximum on the plateau. In fact neither adsorption nor convection maxima appear in the steady state, due to substantial flux a t the surface of the electrode. The heights of the first and second waves in Figure 2 depend linearly on DecI concentration over the entire range investigated. The ratio of the height of the second wave to that of the first is approximately unity (1.09),which is evidence for two one-electron reduction steps. This clear situation is the result of the unique property of microelectrodes: The influence of surface phenomena and following chemical reactions is insignificant under these conditions. Although LePerriese et al. (21) have found conditions in mixed solvent that yield well-shaped normal pulse waves for DecI, this regular behavior is strikingly different from the typical complex response a t a conventional mercury drop electrode (5, 21). Similar behavior is observed for ButI and EtI. Under the conditions of Figure 2, the corresponding calibration plots for the total wave height have slopes ai/aC (nA mM-') = 7.51, 7.09, and 5.78 for EtI, BuI, and DecI, respectively. The diminishing slope of the lines in the sequence EtI, ButI, and DecI corresponds to decreasing values of the respective dif-
5
10
20 25 30 16' C / m o l / d d
15
~~
- 0.5
- 1.5
-2.5
E/ V
Figure 2. LSV curves of DecI in PC containing no supporting electrolyte: CDecI,0.1, 0.2, 0.4, 0.6, 1.0, 1.7, 2.0, 2.8 mM; Pt/Hg microelectrode; Y = 20 mV/s; T = 25 O C . Inset: first wave height (I) and total height of two waves (11) plotted vs CkI. Electrode: 48 pC of Hg on 12.5-pm-radius Pt.
Table I. Diffusion Coefficients and El12Values of Two LSV Reduction Waves of Ethyl, Butyl, and Decyl Iodide in Pure Propylene Carbonate
Et1 ButI DecI
EljZ(1)
E1,,(2)
106D,cm25-l
-2.28 -2.16 -2.03
-2.59 -2.54 -2.48
3.12 2.93 2.34
fusion coefficients. From the limiting LSV currents, diffusion coefficients of alkyl halides could be determined. To do this, the amount of mercury plated on a 12.5-pm-radius Pt disk was set exactly a t 26.8 pC so that an ideal hemisphere could be formed, and experiments were carried out a t a very low scan rate to achieve a steady-state limiting current. We assume the equation describing the limiting steady-state current at this electrode is i, = 2rnFCDr. The diffusion coefficients calculated in this way and the corresponding Ellzvalues are presented in Table I. According to the discussion above, we formulate the steady-state current on the 48-pC electrode as is, = 8.245nFDCr, from which the slope of the calibration curve at steady state is predicted to be (n = 2) 6.21, 5.83, and 4.65 nA/mM for ethyl, butyl, and decyl iodide, respectively. The experimental values are about 17% higher than the predicted steady-state values, in good agreement with the estimate of non-steady-state diffusion for Figure 2. Reverse Pulse Voltammetry. According to Bilewicz and Osteryoung ( I S ) , the electroreduction of alkyl iodides a t mercury electrodes can be described by the reactions RI
+ Hg + eRI
-
RHg'(ads)
+ 2e-
+ I-
R- + I-
The radical RHg' can dimerize to form HgR2 and can react with RI to produce I-. The anion R- can react with proton donors (13, 22) or with reactant:
R- + HD -,RH + DR- + RI R2 + I-
-
ANALYTICAL CHEMISTRY, VOL. 62, NO. 4, FEBRUARY 15, 1990
352
- 2.8
-2.7 - 2.6 I5nA
- 2.L - 2.3
1 8 nA 1
I
I
0.0
~1.0
-2.0
E/ V
Flgure 3. Reverse pulse voltammetry of D e c I in PC containing no supporting electrolyte: C , 0.1, 0.2, 0.3,0.5,1, 2 mM; t,, 50 ms; delay time (at -2.7 V) = 1 s. Electrode: 48 pC of Hg on 12.5-bmradius Pt.
20
aC \
.-
lo - 0.6
I
- 0.2 I
I
5
10
v m 1
15
20
\
1
I
-1.0
Flgure 4. First and second reverse pulse wave heights and their ratio plotted vs tp1'2. The current i , , for the second wave is measured with respect to i , . Other conditions are the same as in Figure 3.
If HD is a water molecule, then D- is hydroxide ion. The species I-, OH-, R- and RHg' potentially can yield anodic waves, but in fact only in the presence of iodide and hydroxide ions is the Hg electrode oxidized at more negative potentials than the background oxidation. The adsorbed radical, RHg'(ads), is also oxidized anodically in the presence of iodide. This situation favors choosing RPV as the appropriate technique for monitoring the reduction of RI. In addition, RPV can be accomplished easily at microelectrodes, since boundary conditions are readily renewed during the waiting periods between pulses (23). Note that because Hg is oxidized during the pulse period as the RPV curve advances, the size of the electrode and the pulse width must be chosen to avoid a significant decrease in the electrode area. Typical R P voltammograms of DecI in pure propylene carbonate are presented in Figure 3. The curves are background corrected. The maximum pulse amplitude is 3.7 V. The shape of the curves and well-developed plateau indicate that the electrode was appropriately polarized during the pulses. The DC current (at -2.7 V) is predicted, using the approximations given above, to give the slope dide/dC = 7.53 nA mM-'. The experimental currents are linear in concentration with slope 7.40 nA mM-'. Two well-formed reverse pulse waves are seen in Figure 3. We deal first with the time dependence of these limiting currents as displayed in Figure 4. The height of the first wave decreases with respect to that of the second a t longer times. This suggests that the first wave is due to the reaction of a
t1.0
0.0
-1.0
-2.0
E/ V
-3.0
Figure 5 . Reverse pulse voltammetry of 2 mM DecI obtained for various delay potentials, as pointed out at the left side of the figure. Conditions are the same as in Figure 3.
species unstable on the time scale of the experiment, probably RHg' (15). Detailed interpretation of the time dependence of the RPV data is not possible. However, a few further observations can be made. At a disk, the time dependence of the reverse pulse current for an uncomplicated process has the form (l/tp1/2- l / ( t p + tg)'12),where t, is the time at the generating potential (here -2.7 V) and t , the pulse time. For the data of Figure 4, a plot of -iRp vs (l/tP1l2- l / ( t p + tg)'I2) is linear with slope 6.22 nA s1lz and intercept -0.10 nA for the second wave. The first wave also exhibits a linear plot, but the substantial negative intercept (-0.8 nA) suggests the same thing as Figure 4,that a coupled chemical reaction is removing the material responsible for the current at a rate commensurate with the experimental time scale. The experimental values for the ratio iDc/(iDc - iRp) from the data of Figure 3 range from 0.332 (2 mM) to 0.417 (0.1 mM) (iDcat -2.7 V, -iRp at +1.0 V). As iDc agrees well with theory, -iRP is increasing with increasing concentration, as found previously for the same general range of time and concentration (15). We speculate that this is caused by the reaction of R- with RI to form Rz + I-. From the analytical point of view the key point is that the reverse pulse wave height is proportional to concentration of alkyl iodide. For the conditions of Figure 4,the slopes of the calibration plots are -aiRp/aC (nA mM-') = 13.12, 13.56, and 14.29 for decyl, butyl, and ethyl iodide, respectively, and the intercepts are 0.01, 0.07, and 0.31 nA, respectively (sr= 0.34, 0.22, and 0.06 nA, respectively; r L 0.999). The amount of iodide produced depends sensitively on impurity levels, and hence the slopes of these curves are sensitive to additions of supporting electrolyte. These results show that the limiting reverse pulse currents can be used analytically in the absence of supporting electrolyte. Figure 5 displays reverse pulse voltammograms for the reduction of decyl iodide a t various potentials. As the potential is made less negative, the limiting reverse pulse current decreases, as expected (15). Furthermore, the height of the second anodic wave decreases with respect to that of the first, and the two waves are no longer distinct. This is consistent with the increasing importance of the radical RHg' at less negative reduction potentials. The prominence of the first anodic wave shown in both Figures 3 and 5 at rather high concentrations and long pulse widths suggests that the lifetime of the radical is much more longer under the present conditions than under the usual conditions employing excess supporting electrolyte. These results demonstrate the feasibility of carrying out not only linear scan but also normal and reverse pulse voltammetry a t platinum-based mercury microelectrodes in solvents of sufficiently high dielectric constant in the absence
Anal. Chem. 1990, 62,353-359
of supporting electrolyte. The spherical segment electrode is well-behaved experimentally even with very large (3.7 V) pulse amplitudes. Furthermore, the limiting cathodic currents agree well with the semiempirical prediction for the spherical segment. The DC and reverse pulse limiting current are also proportional to concentration, which suggests that this experimental approach could be useful for in situ monitoring of the composition of resistive liquids. The quality of the experimental results also suggests that it would be worthwhile to develop the necessary models for studying reaction mechanisms by reverse pulse voltammetry in such systems. In the absence of supporting electrolyte the lifetime of unstable species is extended because the solvent medium is more inert. This can offer significant advantages in the study of complex reaction mechanisms. It may also improve reproducibility of experimental conditions. The voltammetric response in reverse pulse provides a picture of the reactions occurring (as does the response in cyclic voltammetry). The reverse pulse limiting currents, which are unaffected by iR drop, provide the accuracy and relatively simple interpretation characteristic of double potential step chronoamperometry. Particularly in the absence of supporting electrolyte, for which case complicated models are required, it would seem reasonable to focus on analysis of potentialindependent currents. Thus reverse pulse voltammetry appears to be an attractive tool for investigating homogeneous reaction mechanisms under these conditions. LITERATURE CITED (1) Pons, S.; Fleischmann, M. Anal. Chem. 1987, 59, 1391A. (2) URramlcroelectrcdss;Fleischmann, M.,Pons, S., Rolison, D., Schmidt, P. P., Eds.; Datatech Science: Morganton, NC, 1987. Wightman, R. M. Anal. Chem. 1984, 56. 524. (3) Howell, J. 0.;
353
Bond, A. M.; Flelschmann. M.; Robinson, J. J. Electroanal. Chem. Interfacial Electrochem. 1984, 168,299. Bond, A. M.; Fleischmann, M.; Robinson, J. J. Electroanal. Chem. Interfacial Electrochem. 1984, 172, 11. Ciszkowska, M.;Stojek, 2. J. Electroanal. Chem. Interfacial Electrochem. 1986, 213, 189. Stojek, 2 . ; Osteryoung, J. Anal. Chem. 1988, 6 0 , 131. Bond, A. M.; Lay, P. A. J. Electroanal. Chem. Interfacial Electrochem. 1988, 199,285. Oldham, K. B. J. Electroanal. Chem. Interfacial Electrochem. 1988, 250, 1. Bond, A. M.; Fleischmann, M.; Robinson, J. J. Nectroanal. Chem. Interfacial Electrochem. 1984, 172, 11. Bruckenstein, S . Anal. Chem. 1987, 59,2098. Bixler, J. W.; Bond, A. M. Anal. Chem. 1986, 58,2859. Ghoroghchian, J.; Sarfarazi, F.; Dibble, T.; Cassidy, J.; Smith, J. J.; Russel, A.; Dunmore, G.; Fleischmann, M.; Pons, S. Anal. Chem. 1988. 58, 2278. Bond, A. M.; Henderson, T. L. E.; Thormann, W. J. f h y s . Chem. lg88, 90,2911. Bilewicz, R.; Osteryoung, J. J. Electroanal. Chem. Interfacial Electrochem. 1987, 226,27. Huang, H.J.; He, P.; Fauikner, L. R. Anal. Chem. 1988, 58, 2889. Stojek, 2 . ; Osteryoung, J. Anal. Chem. 1989, 61, 1305-1308. Pena, M. J.; Fleischmann, M.; Garrard, N. J. Electroanal. Chem. Interfacial Electrochem. 1987, 220,31. Lines, R.; Parker, V. D. Acta Chem. Scan. Ser. 6 1977, 31,369. Aoki, K.; Akimoto, K.; Tokuda. K.; Matsuda, H.; Osteryoung, J. J. Electroanal. Chem. Interfacial Electrochem. 1984, 171, 219-230. La Perriere, P. M.; Carroll, W. F., Jr.; Willett, 8. C.; Torp, E. C.; Peters, D. G. J. Am. Chem. SOC. 1979, 101, 7561. Andrieux, C. P.; Gilardo, J.; Saveant, J. M.; Su, K. B. J. Am. Chem. SOC. 1988, 108, 638. Sinru, L.; Osteryoung, J.; O'Dea, J. J.; Osteryoung, R. A. Anal. Chem. 1988, 60, 1135.
RECEIVED for review September 12,1989. Accepted November 14,1989. This work was supported in part by the U.S. Office of Naval Research and by Grant 01.17.04.01 from the Polish government. Preliminary results were presented at Euroanalysis VI, Paris, 1987.
Gas Sorption to Plasma-Polymerized Copper Phthalocyanine Film Formed on a Piezoelectric Crystal Shigeru Kurosawa, Naoki Kamo,* Daijyu Matsui, and Yonosuke Kobatake Faculty of Pharmaceutical Sciences, Hokkaido University, Sapporo 060, J a p a n
Copper phthalocyanine was plasma-polymerized on plezoelectrlc quartz crystals. The film formed was very stable and resistant to chemical and physlcal treatment. Various spectroscopic data suggested that the essential properties of phthalocyanine remained after the polymerization. A variety of organk chemicals were examined to measure the sorption to the flim from the gas phase (In the presence of air). Adsorption of gases to the fllm decreased the osclllatlng frequency. Sensltlvity was defined as the frequency decrease per 1 part per mlilion (ppm) of a certaln gas, and the values ranged from 670 to 0.001 Hr/ppm. Chemicals showing high affinity are plane molecules with conjugate double bonds and some polar substituting groups or are higher alcohols, such as vanlllln, benzoic ecld, aniline, nitrobenzene, phenol, n-octyl alcohol, 1-nonanoi, n-decyl alcohol, benzyl alcohol, DL-camphor, &ionone, naphthalene, and anthracene. The film on the piezoelectric crystals was proved to have a life as long as 60 days. The sensltlvity to varlous Chemicals was dependent on a central metal of phthalocyanlne.
* Author to whom correspondence should be addressed.
INTRODUCTION The frequency of vibration of an oscillating piezoelectric crystal is decreased by adsorption of a foreign substance onto its surface. Sauerbrey (I) derived an equation describing the relationship between frequency decrease and the weight of substances attached to a surface. The detection limit of a 9-MHz oscillating crystal has been estimated to be as low as lo+ g. By coating the crystal with some absorbent substances, it is possible to construct gas sensors. Since King (2) reported the use of piezoelectric crystals for gas sensors, their features of low detection limit, broad application range, and continuous operation mode have attracted interest (3-14). Phthalocyanines (Pc) are known to be good adsorbents for various chemicals. In addition, they work as a p-type semiconductor in the presence of oxygen, and adsorption of electronegative gases changes their electric resistance appreciably (15-18). Using this interesting nature of Pc, Barendsz, et al. (19,20),Ricco et d. (21),and Roberts et al. (22) fabricated a surface acoustic wave (SAW) device covered with a thin Pc film and succeeded in obtaining a highly selective and sensitive response to halogen gases and NOz gases. The device was operated at 150 "C,and gradual loss of the film
0003-2700/90/0362-0353$02.50/00 1990 American Chemical Society
